25

The Preparatioga arzd Polymerisatio .n of -Methyl Vimyl Ketone.

[I9431
9.

The Preparation and Polymerisation of Methyl Vinyl Ketone.

By THEODORE
WHITE and ROBERTW. HAWARD.

Methyl vinyl ketone is of importance in view of its capacity for polymerisation to a thermoplastic resin.
A detailed study has been made of the acetone-formaldehyde condensation with a view to establish the optimum
conditions for the production of 3-ketobutanol, from which methyl vinyl ketone is obtained by dehydration.
The condensation is complex and its various products are discussed and characterised. The preparation of pure
methyl vinyl ketone has been studied, and its rate of polymerisation measured in a series of solvents. Both
the high rate of polymerisation obtained in precipitants and the discrepancies observed in the kinetics of the
polymerisation in benzene can be explained by the assumption that chain termination is retarded in liquids
which do not dissolve the polymer.

THEcondensation of acetone and formaldehyde to 3-ketobutanol and the dehydration of the latter to methyl
vinyl ketone were first reported by Merling and Kohler (B.P., 1909, 19087; J . SOC.Chem. Ind., 1910, 29, 1037).
Later work, summarised by Morgan, Megson, and Pepper (Chern. and I n d . , 1938, 57, 885), estabfished that the
condensation was a complex process and, with the present contribution, suggests that the reactions proceeding
on condensation and subsequent distillation can be represented as in Fig. 1.
FIG.1.

CH,
- H,O (Distn.) CHz(OH)
- (HpO + CH,O)
\\C.COMe
+------)CHCOi\le
- ------ f CH,:CH*COMe
OHCI-T,/
CH, (OH)
(Distn*)
n!tethyl vinyl ketone.
Methylene-3-ketobutanol.
Bishydroxymethylacetone.

I ..+
1

H 2 0 (Distn.
with catalysts.)

0-CH
SET,<

')CH-COMe +

3CH,O

J(cH3),C0 ~

1 : 3-Dioxanyl-5
0-CH, methyl ketone.
&(CH')aCo

+-

I
-+ CH,(OH)CH,*COMe

CH,O

CH,*COCH,
e

'

,
CH,*COCH,*CMe,*OH

Diacetonyl alcohol.
I

\1-H20 (Distn. with catalysts)

1 : 3-DioxanyI-5 isobutenyl

CH,-CO*CH:CMe,
Mesityl oxide.

ketone.

Resin intermediates.

- (Ha0 -t
CH,O)
(Distn.)

Insoluble cross-linked resins.

Little attention has been given t o the isolation and purification of each component or t o the manner in
which variation of the condensation conditions affects the yield of 3-ketobutanol. It has, however, been
claimed by Morgan et.aZ. (Zoc. cit.) and Rutovsky, Berlin, and Zabyrina ( J . Gen. Chem. Russia, 1941, 11, 550)
that 3-ketobutanol enolises to form a terminal methylene group,
CH,(OH) *CH,COCH,

--+CH,(OH)CH,C(OH)XH,,

and then polymerises under the influence of light or peroxides, a phenomenon not observed with other enols.
The necessity of obtaining pure methyl vinyl ketone before studying its polymerisation led t o the present
study of the acetone-formaldehyde condensation and t o the elucidation of discrepancies in the earlier work.
By methods such as those of Fenske, Tonsberg, Quiggle, and Cryder ( I n d . Eng. Chem., 1936, 28, 644; 1937,
29, 957) and Podbielniak ( I n d . Eng. C h e w Anal., 1933, 5, 176), the condensation product left after removal
of excess of acetone was thoroughly fractionated, thus giving details of the amounts of each component present
and providing the products in a very pure state. This gave accurate physical constants for each product
and an explanation of the polymerisation of 3-ketobutanol.
Earlier workers found that most of the condensation product resinified on distillation. Pressure variation,
control of pH, and elimination of oxygen and peroxides did not prevent this, but by distillation in the presence
of a high-boiling ester (butyl phthalate) it was found that the volatile components distilled without decomposition and the non-volatile products were precipitated unresinified from the cold ester. High-boiling
paraffins did not show this effect, and the reason for the normal occurrence of resinification and its prevention
by butyl phthalate remains obscure. The procedure permitted isolation of 70% of the crude product as
volatile substances, which were refractionated without decomposition.
Application of this discovery established that a maximum yield of 3-ketobutanol was obtained by refluxing
a 4 : 1 molar ratio of acetone : paraformaldehyde a t p H 8.3-8.5 for a time just sufficient to permit complete
reaction of the paraformaldehyde, the type and effect of variation of the condensation conditions being detailed
in the Experimental section.
The product obtained under these conditions contained, after removal of the excess of acetone, 30% of
non-volatile products, which were not investigated in detail, the volatile portion consisting of : (1) acetone ;
(2) water (together 9-10%);
(3) diacetonyl alcohol, b. p. 63"/12 mm. (4-5%);
( 4 ) 3-ketobutanol, b. p.
70-71"/12 mm. (27-28%) ; (5) methyleneketobutanol, b. p. 82-83"/12 mm. ( 1 6 1 5 % ) ; ( 6 ) 1 : 3-dioxanyl-5
isobutenyl ketone, b. p. 90-92"/12 mm. (lO-llyo) ; (7) 1 : 3-diosanyl-5 methyl ketone, b. p. 16Oo/l2 mm.
(3-4Y0).
"

"

26

White aizd Haward: The Preparation aqzd

Fractionation other than by the advanced technique cited gave a main, superficially pure fraction, b. p.
90-92"/12 mm.-the " 3-ketobutanol " of earlier workers. This was found to be a mixture of components
(3)-(6), a fact which explains the supposed polymerisation of 3-ketobutanol. We have been unable t o
polymerise pure 3-ketobutanol of b. p. 70-71"/12 mm. and have no evidence of its enolisation. On the other
hand, methyleneketobutanol, b. p. 82-83"/12 mm., polymerised with both light and peroxides, and its presence
to the extent of some 30% in the mixture of b. p. 90-92"/12 mm. is sufficient t o explain the 30% maximum
yield of polymer obtained by Rutovsky et aZ. (Zoc. cit.) from their 3-ketobutanol and the similar polymerisation
recorded by Morgan et al. (Zoc. cit.).
Of the components (1)-(7), the occurrence of diacetonyl alcohol and 1 : 3-dioxanyl-5 isobutenyl ketone
has not previously been noted. The formulation of the latter rests upon analysis and cryoscopic molecularweight determination ( M , 165). Bromine absorption showed the presence of one double bond and a molecular
weight of 160-170 ; and the production of acetone on oxidation with alkaline permanganate located this double
bond in a terminal isopropylidene group. Hydroxylic oxygen was found to be absent, but although, like mesityl
oxide and bishydroxymethylacetone, the substance did not give a bisulphite derivative, oxime, or phenylhydrazone, yet the presence of a keto-group was established by the formation of a 2 : 4-dinitrophenylhydrazone.
With hot dilute acids formaldehyde was produced, indicating the presence of the remaining two oxygen atoms
in a methylenedioxy-group. These facts can be interpreted by more than one formula, but all require the
presence in the molecule of a substituted 1 : 3-dioxan ring. Morgan and Holmes (J., 1932, 2667) established
the presence of an allied substance, 1 : 3FIG.2.
dioxanyl-5 methyl ketone, in this same conThe polymerisate'on of methyl vinyl ketone in benzejte solution at 25" densation, and M~~~~~ and Griffith ( J., 1937,
with approximately 1% of benzoyl peroxide.
841) found a similar type of product in the
n
80
methyl ethyl ketone-formaldehyde condensation product. The formulation of product
(6) as 1 : 3-dioxanyl-5 isobutenyl ketone is
6-4
tentative, but the established presence in
C'
the condensation mixture of 1 : 3-dioxanyl-5
.o 60
methyl ketone, from which i t can easily arise
4
by condensation with acetone (cf. the form.2
ation of diacetonylacetone), makes this the
$
F
most likely of the alternatives.
240
The preparation of methyl vinyl ketone by
Q
dehydration of pure 3-ketobutanol and of the
%
crude condensation mixture was also studied,
cl
the results explaining why its b. p.'s have been
quoted as 74-87"/760 mm. Dehydration of
Q 20
3-ketobutanol gave a constant-boiling methyl
vinyl ketone-water mixture of b. p. 75"/760
mm. Dehydration of the crude reaction
product gave ternary acetone-methyl vinyl
0
400 ketone-water mixtures, which distilled as
0
100
200
300
pseudo-constant-boiling fractions a t any temTime. hours.
perature from 69" to 94"/760 mm. according to
the type and amount of dehydrating agent used. Refractionation always gave a ternary constant-boiling main
fraction, b. p. 73-74'/760
mm., containing 42.5% of acetone, 42.5% of methyl vinyl ketone, and 15% of water,
from which pure methyl vinyl ketone, b. p. 81.4"/760 mm. (corr.),%dii:0.842, d6'1-4120, was obtained by
adding acetic anhydride to eliminate the water, the resultant acetone-methyl vinyl ketone-acetic acid
mixture being readily separated into its pure components by a Fenske column. The methyl vinyl ketone so
prepared was used in the following study of its polymerisation.
Polymerisation of Methyl Vinyl Ketone.-The pure ketone polymerised to a solid d 1.12, i.e., with a asyo
reduction in volume. Such volume changes have frequently been made a measure of the degree of polymerisation (e.g., Bolland, PYOC.
Roy. Soc., 1941, A , 178,24), and the method was used throughout this work, consistent results being obtained. The ketone polymerised slowly in the presence of benzoyl peroxide a t room
temperature and more quickly a t 60". Oxygen appeared to cause an induction period if there was less than
0.1% of peroxide present but the effect was not investigated in detail. The dilatometric method was not
convenient for investigation of the undiluted monomer, since the polymer-monomer solution became nearly
solid when only 20-30%
of the monomer had polymerised, and the following results were all obtained in
solvents.
P d y n z e r i s a t i o n in Benzene.-Polymerisation-time
curves for benzene solutions of various concentrations,
each containing approximately 1% of catalyst] are given in Fig. 2. For concentrations below 30% the curves
are smooth and convex, as if the reactions were following first- or second-order kinetics. Table I illustrates
for three concentrations of ketone the result of applying the usual relationships to these curves. The results
are remarkable in that, if constancy of the velocity constant ( k ) during any individual reaction is taken
as the criterion, the reaction is of first or 3/2 order, whereas if constancy of the velocity constant for various
"

'I

"

"

9)

"

"

"

"

119431

27

PoZynzerisation of IWethyZ Vi9ayZ Ketone.

initial concentrations is the criterion, then the result is intermediate between a 312 and a second-order reaction.
Clearly, the reaction cannot be assigned to any definite order.

TABLEI.
Velocity Consta9zts for the Polymerisation of Methyl Vinyl Ketone iiz Benzene (first-, 312-, and second-order
constants).
Ketone, 28.7y0 by vol. = 3.48 g.-mols./l. ; benzoyl peroxide, 0.83% = 3.4 X lo-, g.-mol./l.
11.
19.
30.
47.
71.
95.
119.
Time, hrs.
lo4 k, ........................
41
39
40
40
37
31
33
26
28
26
27
26
lo4 k 3 ! , ........................
26
26
30
33
33
36
36
104
........................
29
25
Ketone, 9.6% by vol. = 1.14 g.-mols./l. ; benzoyl peroxide, 1.0% = 4.1 x lo-, g.-mol./l.
Time, hr s.
24.
48.
72.
96.
120.
144.
168.
16
15
16
15
15
15
lo4 k ,
........................
12.5
19
18
19
19
17
18
16
lo4 k3/, ........................
38
38
40
41
32
34
31
lo4 k , ........................
Ketone, 1.9% by vol. = 0.23 g.-mol./l. ; benzoyl peroxide 0.99% = 4.1 X lo-, g.-mol./l.
880.
1,296
600.
Time, hrs.
216.
384.
4.2
3.8
4.4
4.3
104 K , ........................
4.3
13
12
11
11
13
1 0 4 k 3 / 2 ........................
67
71
62
lo4 k , ........................
44
53

143.
34
28
40

K,

192.
15
20
43

This point is further emphasised by Table 11, which gives a complete series of constants for the range
0*49-36.7% of methyl vinyl ketone (by vol.), the most satisfactory constant for each individual reaction
being italicised. The results a t the lowest concentrations are the least accurate, since the volume changes
measured were \7ery small. The results will not be discussed here, but it may be noted that a similar result
was recorded for the polymerisation of vinyl acetate in toluene solution by Cuthbertson, Gee, and Rideal
(Proc. Roy. Soc., 1939, A , 170, 315).
TABLE11.

Velocity Comtants for the Polymerisation of Methyl Vinyl Ketone in Benzene at 0.49-36.7%
(0-059-4.4 g.-mol./l.).
Approximately 1yo Catalyst throughout.
1.14.
0.55.
0.059.
0.23.
15
9.0
104 K, ..............................
2.8
4.2
18
16
12
104k312 ..............................
15
38
70
130
60
lo4 k , ..............................
Concn., g.-mols. 11.

3.48.
37
27
40

by V O ~ .

4.4.
40
29
39

FIG.3.
Iiztrinsic viscosity of methy1 viizyl ketone polymers formed in various solvents at 26" with approximately
1 yo of benzoyl peroxide.

c
A cyc/oHexane.
~ 3 1isoPropy/ ether.
0 Benzene.
a Acetone.

Changes in Ipitrinsic Viscosity.-Fig. 3 depicts the intrinsic viscosities ( q i ) (cf. Gee, Trans. Faraday SOC.,
1940, 36, 1167) of methyl vinyl ketone polymers formed in benzene and other solvents in the presence of approx.
1% of benzoyl peroxide. The results show that for benzene solutions the final viscosity is approximately

White nfld Haward : The PreParatioPz and

28

proportional to the initial concentration of ketone. Hence it may be concluded that this variation in chain
length will, of itself, introduce a factor proportional to the concentration of the ketone into any kinetic equation
and therefore will raise the order by unity. If the evidence given above were taken as implying a first-order
reaction, all the other processes would have t o be zero order as far as the ketone concentration is concerned.

TABLE111.
The Effect of Catalyst Concentration on the Polymerisation of Methyl Vinyl Ketone.
3-2
3.5
Ketone, g.-mols./l. ....................................
3-5
3.7
Catalyst (C,), g.-mol. /l. ...........................
0.034
0.0145
0.0062
0.0025
108 K, ................................................... 3.7
20
103 k , x qdc,, .......................................
Intrinsic viscosity .................................... 1.08
Intrinsic viscosity x d G ........................ 0.20

2.1
18
1.42
0.17

1.9
24
1-83
0.14

1.2
23
3.0
0.16

3.9
0.0013
0.92
26
3.1
0.1 1

The Effect of Catalyst Concentration.-These results suggested that any investigation of the effect of catalyst
concentration should be carried out a t a single concentration of ketone, whereby the effects due to the uncertain
order of the reaction would be largely eliminated. In Table I11 the first-order constant ( k , ) is given, in these
FIG.4.
Polymerisation of methyl vi~zylketone in various solvents at 25" with approximately 1% of bemoyl peroxide.

circumstances, purely as an estimate of the rate of reaction. The results show that the rate of reaction is
proportional t o
as is usually the case for these catalysed polymerisations (cf. Cuthbertson et al., ZOG. cit. ;
Schultz and Husemann, 2. physikal. Chem., 1938, B , 39, 246; Price and Kell, J . Amer. Chern. SOC.,1941, 68,
798), and it may be concluded that the reaction chain is terminated by a process of mutual saturation. The
fact that the product, chain length x
is much less constant and falls a t low concentrations of the catalyst
suggests that the reaction chain and the polymer chain may not always be synonymous and that a process of
chain transfer such as that postulated by Kamenskaya and Medvedev (Acta Physicochim., 1940, 13, 565),
occurs.
The Eflect of Variation of the So,?vent.-Fig. 4 depicts the polymerisation of methyl vinyl ketone in a series
of solvents ; it is obvious that the rates of reaction differwidely, particularly in the later stages. This difference
appears t o be correlated with precipitation of the polymer in the same way as that observed for methyl methacrylate by Norrish and Smith (Nature, 1942, 150, 336). In both cyclohexane and isopropyl ether only the
monomer is soluble, the polymer being precipitated as soon as any quantity is formed, whereas both acetone
and ethylene dichloride are good solvents for the polymer. Reference to Fig. 3 shows that the solvents which
give the highest reaction rates also give the longest chain lengths, but the difference in the rate of reaction is
greater than that in molecular weight.
In view of the complex nature of the curves, no attempt has been made to analyse the reaction in cyclohexane or isopropyl ether, but the reactions in ethylene dichloride and in acetone appear to give satisfactory

Y'G,

<,

Polymerisatio.ta of Methyl Vinyl Ketoae.

El9431

29

3/2- and second-order constants, respectively, as may be seen in Table IV. No allowance has been made for
decomposition of the benzoyl peroxide in any of these reactions. It is certainly slow a t 25", but it could easily
be fast enough to affect the kinetics of the longest reactions (50 days). The fact, however, that consistent
kinetics are given in ethylene dichloride after this time suggests that the decomposition cannot be large and,
in any case, the inconsistencies obtained with benzene cannot be attributed to this cause.
TABLEIV.
Velocity Constants f o r the Polymerisation of Methyl V i n y l Ketone in Acetone and in Ethylene Dichloride.
[Constants cwrected f o r small deviations of C, from 0.041 g.-mol./I. ( l % )C,
, being assumed low.]
Ketone, g.-mol. 11. ........................
lo3 k ,
.......................................

Time, hrs.
1CP kslz

.........

...............

Acetone.
1.0
1.5

2.7
2.5

Ethylene dichloride.
Methyl vinyl ketone = 0.16 g.-mol./l.
888
1296
600
384
120
316
3.3
2.5
2.3
2-5
3.0
2.4

3.4
1.9

4-75
1.9

Methyl vinyl ketone = 3.0 g.-mols./l.

19
43
67
91
2.0
1.9
2.0
3.0

Discussion.-The results in different solvents are in agreement with those of Norrish and Smith (Zoc. czt.)
and can be explained on similar lines-that precipitation reduces the rate of chain termination and hence
increases the rate of reaction. The fact that the polymer chain length is not equivalently increased shows that
the interaction of two active polymer chains plays a more important part in terminating the reaction chains
than the polymer chains, i.e., that chain transfer occurs.
This precipitation theory can also be extended t o cover the irregularities in the kinetics of polymerisation
in benzene, which is a poor solvent for the polymer. Here the accumulation of micellar aggregates may
cause a similar rise in the rate of reaction as the polymer is formed and so cause a single reaction to appear of
lower order than it really is. The theory would, however, appear to predict that the rate of polymkrisation
would be faster in a viscous solution of pure polymer than in a pure solvent. To test this, an experiment was
made in acetone, since the polymer could not be dissolved in benzene, and it was found that in a 3% solution
of polymer of -qi = 2-6 the rate of reaction was the same as in the pure solvent. An explanation of this apparent
discrepancy may be that the increase in viscosity when a small amount of polymer is present in a good solvent
does not affect the agitation of the different parts of a single chain, which is the most important factor in the
termination process, although it may reduce the rate of diffusion of the centre of gravity of the chain. It
seems probable, however, that some factor of the viscosity type does play a part in the reaction owing to the
relative ease with which simple monomeric substances polymerise to hard materials with very small internal
monomeric concentrations.
Although the kinetics of the polymerisation cannot be ascertained with certainty, it seems probable that a
bimolecular reaction takes place in benzene and acetone which can be consistently explained as :

Initiation : M Catalyst >

Complex
Complex + M ___p. Chain
Catalyst
Termination : P,+ P,+ +P, + P, (dead polymer)
Continuation .- P, + M >
-,
P,,,; rate cC [P+][M]
i.e., chain length is proportional to [MI.

Together, these give the rate of reaction as proportional to G [ M I 2 .

The relation of chain length to concentration is, however, an uncertain factor which may account for the
3/2 order in the case of ethylene dichloride. Alternatively, it may be possible for this solvent to play the
part of one of the monomer molecules in the initiation mechanism.
The results do not justify more detailed attention to the reaction mechanisms.
EXPERIMENTAL.
Fractionation.-Fractionations
were carried out with Fenske columns (Zoc. cit.) and a total-condensation still-head,
using a 10/1 reflux ratio and a distillation rate of 1-3 g./min. The column for atmospheric-pressure distillations had an
internal diameter of 14 mm. and was packed for 50 cm. with 3 mm. diameter single helices made from 26-gauge nickel
wire. Reduced pressure distillations were conducted with a similar type of column of 25 mm. internal diameter packed
for 45 cm. with 6 mm. helices. The Podbielniak (loc. cit.) type of column was used for comparative purposes and gave
identical results.
3-KetobutanoZ.-3 L. of acetone, 315 g. of paraformaldehyde, and 20 ml. of 2~-methyl-alcoholicpotash (pH = 8.38.5) were refluxed with mild stirring until 4 drops of the reaction mixture gave no reaction with 2 ml. of Tollens's reagent
(ca. 5 mins.), indicating complete reaction of the paraformaldehyde. The reaction mixture was then neutralised with
B~-hydrogenchloride in acetone, and the excess of acetone distilled off on a water-bath until the distillation ceased (recovery = 2365 ml.), leaving the crude condensation product (800 g.). 500 M1. of butyl phthalate were added to this
product, and the mixture distilled under reduced pressure, volatile products distilling up to 180"/15 mm. (575 g.). The
remaining non-volatile components of the crude product were precipitated when the butyl phthalate residues cooled.
The distillate was fractionated with the specified Fenske column, the third distinct fraction being 3-ketobutanol,
b. p. 70-71"/12 mm.,
1.029, nls0 1.435; yield 215 g., 23% (Found : C, 54-2; H, 9.0; OH, 18-9. Calc. for C,H,Op :
C, 54.5; H, 9.1 ; OH, 19.3%). It did not polymerise with peroxides or U.V. light or on exposure t o ordinary light for

e:

The Pyepayation and Polymerisation of Methyl Vinyl Ketone.

30

12 months. Contrary t o the statement of earlier workers, it boiled without decomposition (b. p. 182') a t 760 nim.
It gave a bisulphite derivative, resinified with alkalis, and was dehydrated to methyl vinyl ketone by numerous
dehydrating agents.
Composition of the Crude Condensation Product.-Many samples of crude product prepared as above were fractionated
thoroughly, only slight variations being noted. A typical 800-g. sample distilled from butyl phthalate gave 226 g. of
non-volatile precipitate and 575 g. of volatile products. Fractionation of the latter gave the fractions : (i) b. p. 1545"/50-20 mm., 80 g. (9--10%), i.e., water and acetone; (ii) b. p. 63'/12 mm., d6*
1.4225, 33 g. (4-5%) ; (iii) b. p.
70-71"/12 mm., 215 g. (27-28%), 3-ketobutanol (cf. above) ; (iv) b. p. 82-83"/12 mm., dz;: 1-06, n15" 1.4675, 110 g.
(14-15%) ;
b. p. 90-91"/12 mm., d5'1-4720, 80 g. ( l O - l l % )
: (vi) b. p. 16Oo/l2mm., 25 g. (3-4%).
Fraction (11) was diacetonyl alcohol, formed by condensation of two acetone molecules; it had b. p. 164'/760 mm.,
and with acids or iodine gave mesityl oxide, b. p. 129"/760 mm., nl" 1.4435, with the smell of peppermint and other
characteristics of this compound. Fraction (iv) consisted of methylene-3-ketobutanol ; it absorbed bromine immediately
(94% of the theoretical quantity) and had the characteristic properties of an ethylenic ketol, giving a bisulphite derivative
and an acetate, b. p. 95"/25 mm., nI5* 1.4530, and conformed with the product finally described b y Gault and German
(Cornpt. rend., 1936, 203, 514) as having b. p. 84-88"/13 mm. It polymerised slowly under the influence of light or
peroxides to a product of low molecular weight (Found : C, 59-7 ; H, 8 . 5 ; OH, 18.0. Calc. for C5H,02 : C, 59.4; H,
8.9; OH, 18.0yo). The acetate (bromine absorption = 97% of the theoretical) polymerised similarly but much more
slowly. Fraction (v),believed t o be 1 : 3-dioxanyl-5 isobutenyl ketone, gave a 2 : 4-dinitrophenylhydrazone but no oxime
or bisulphite derivative. It could not be acetylated. Analysis indicated the empirical formula C3Hd50,and bromine
absorption (one double bond being assumed) a molecular weight of 177 [Found: C, 63-7; H, 8 .0 ; M (Rast), 165.
C,HI,O, requires C, 63.5; H, 8.2%; M , 1701. Oxidation with cold alkaline permanganate gave acetone (somewhat
less than 1 equiv.), identified after distillation from the reaction mixture as its 2 : 4-dinitrophenylhydrazone and estimated
by the iodoform reaction. Treatment with hot 2~-hydrochloricacid split off formaldehyde (1 equiv.), estimated as its
phloroglucinol condensation product (Morgan and Griffith, Zoc. cit.). The occurrence of this fraction has not been previously noted. Fraction (vi) corresponded in such details as were available with the 1 : 3-dioxanyl-5 methyl ketone
described by Morgan et al. (loc. cit.) and previously designated " a methylene dioxy derivative."
The non-volatile product (Fraction vii) , precipitated on cooling the butyl phthalate residues, was isolated by adding
an equal volume of ether and filtering off the product, which was washed with ether and dried, giving a pale yellow
This product resinified on distillation even at 0.0001 mm. ; it was soluble in
amorphous powder ; yield 225 g. (28-29y0).
acetone, alcohol, chloroform, and ethyl acetate, but not in ether or light petroleum. Fractional precipitation indicated
it to be non-homogeneous, and viscosity determinations on solutions of the material indicated a low average molecular
weight. I t was not further investigated.
Condeizsation Conditions and the Yield of 3-KetobutanoZ.-The conditions giving a maximum yield of 3-ketobutanol
have already been described. The use of different acetone-paraformaldehyde ratios, of other condensation temperatures,
p H values, times of condensation, or of gradual addition of the paraformaldehyde, all gave lower yields of 3-ketobutanol
due to incomplete condensation or excessive resin intermediate formation according t o the factor involved. Paraformaldehyde was preferable t o formalin, reagent purity (within normal limits) had little effect, and the alkalis were the
best catalysts for the condensation. The adverse effect of varying one factor could not be completely compensated by
varying another. Details of all the experiments carried out cannot be given, but Table V gives the results of a number
of comparative experiments designed to elucidate the effect of variation of the condensation conditions upon the yield of
3-ketobutanol. All the condensations were carried out with 315 g. of paraformaldehyde and, except where stated, a
4 : 1 molar ratio of acetone : paraformaldehyde-the
reactions being conducted a t the b. p. of the mixture with
P~-methyl-akoholicpotash as catalyst except where otherwise indicated.

(v)

TABLEV.
Iliustvating the Effect of Condensation Conditions on the Yield of 3-Ketobutanol.
Molar ratio,
COMe, / (CH20)
4/1
19

,*

,,

~N-M~OH-KOH,
ml.
PH.
9
7.5
11
7-8
15
8-0
17
8.3
20
8.5
22.3
8.8
25
9.2
40
10.0
9
7.5
20
8.6
40
10.0
20
9.3
20
8-9
20
8.5
20
8.0
5 ml. C,H,K
20 g. Ca(OH),
7.6
60 g. K,CO,
7.5

Time of condensation (mins.).

5
,>

,I

66

Remarks.
Yield, g.
112 * Incomplete condensation
183 7
194
Maximum efficiency a t p H
212
8-3-86;
higher p H
320 * > lowers yield by increas169
1 ing resin intermediate
147
formation
122 * I
Compare with 5-min. re-

5
I,

360
120
120

:ii0
180

90

p H changes complicate
but do not disguise
J trend of the results
No condensation
Complete condensation i n
both cases

Methyl Vinyl Ketone.-The crude product (800 g.) from a condensation such as that above was heated with 80 g. of
concentrated phosphoric acid, the distillate being fractionated through a Fenske column as it was produced ; the main
fraction distilled at 73-74"/760 mm. ; nI6"1.39-1-406; yield 200 g. Refractionation gave back the same product-a
ternary acetone-methyl vinyl ketone-water mixture in the ratio 3 : 3 : 1. This was treated with an equal weight of
acetic anhydride, kept at room temperature for 24 hours t o remove water, and then refractionated, giving sharp acetone,
methyl vinyl ketone, and acetic acid fractions. The methyl vinyl ketone so obtained was a colourless, lachrymatory
liquid, b. p. 81.4"/760 mm. (corr.), d5'
1.4120, d;$0.842; yield 95 g., ie., 54% on the 3-ketobutanol in the crude product
or 13% overall yield.
Dehydration of pure 3-ketobutanol in the above manner gave a methyl vinyl ketone-water binary mixture, b. p.
75"/760 mm., which again required acetic anhydride treatment before the pure ketone could be isolated. The product,
again produced in a 55% yield, was identical with that above. No further purification of this product was possible

[I9431

31

Bennett, Pritchard, and Simonsen-: Di~~itrodibenza?throne.

nor were any impurities detected. The pure ketone polymerised readily t o a hard colourless resin which absorbed water
readily to give the soft rubbery polymer generally described.
The Effect of varying the Type and Amount of Dehydrating Agent.-A number of dehydrating agents were used to convert 3-ketobutanol and the crude condensation product into methyl vinyl ketone. Each produced a main fraction
with its own characteristic constants, the product in all cases giving the same ternary mixture of b. p. 73-74"/760 mm.,
715'1-39-1.4006, on refractionation. Some of the results are listed below, dehydration being combined with fractionation
through a Fenske column in each case. For routine work, 10% phosphoric acid was added, since it gave the product
closest to that given by refractionation. The yield of ketone varied but slightly with the dehydrating agent used.
Dehydrating
Proporagent.
tion, yo.
H,PO, ............
1
............
10
HjPO, + NaCl
1 +7
CaC1, ............
7

B. p. of
product.
89-92'
73-74
72-74-4
92-94

n.
1.399 -1.3996
1.39 -1-4006
1.3998-1.4048
1.3821-1.3954

Dehydrating
Proporagent.
tion, yo.
HC1 ...............
1
HCl
NaCl ... 1
7
HC1
CaC1, ... 1
7

+
+

+
+

B. p. of
product.
70-73
71-74
68-71

1P.

1*3882- 1a4004
1-3901-1.4030
1.3825-1-4008

Condeizsation Conditions and the Yield of Methyl Vinyl Ketone.-The yield of ketone varied with the condensation
conditions in the same way as did that of 3-ketobutanol, the data below showing the lower yield of product and higher
pH required on carrying out the reaction at room temperature instead of a t the b. p.-the reaction being carried out with
3 1. of acetone and 315 g. of paraformaldehyde.
2s-MeOH-KOH. ml. .....................
7.5
12
a t 25' ............... incomplete condensation
Of
g ' ( a t b. p. ...............
20
78
Technique of investigating the Polymerisation of Methyl Vinyl Ketone.The monomer was fractionated in nitrogen into flask A of the apparatus
shown in Fig. 5, samples of b. p. 80.5-82" only being used. Flask A was
then evacuated, the liquid being allowed t o boil freely to carry away any
dissolved gases in a stream of vapour. The Iiquid was then distilled into
the condenser B so that it ran down into the calibrated tube C which already
contained a weighed amount of catalyst and in which it could be sealed off.
When dilute solutions were being investigated, C in some cases took the
form of a thermometer with capillary and bulb. Before use, C was cleaned
with chromic acid, acetone, and distilled water, and dried. Solvents were
introduced when necessary from the flask D . The charged tube C was
finally sealed, detached, and then attached to a laminated glass scale and
immersed in a thermostat at 25". Illumination was from behind the scale,
and the readings took the form of observation of the downward movement of
the meniscus in C.
Subsequent work on other monomers showed that the vacuum distillation
of the monomer from ,4 directly into the reaction tube was the most essential
precaution for consistent results.
Solvents.-The
solvents used were : Benzene, B.D.H., extra pure for
molecular-weight determinations; cyclohexane, B.D.H., spectroscopically
pure ; acetone, B.D.H., purified by sodium iodide ; ethylene dichloride,
purified by the method of Proskauer and Weissberger (" Organic Solvents,"
p. 157), b. p. 81.45-81.55" ; isopropyl ether, technical, redistilled.

16
30
83

25
72
61

30
60
57
FIG.5 .

yar

The authors express their gratitude to Messrs. Colmore Adhesives Ltd. for
permission to publish these results.
COLMORE
ADHESIVES
LTD.,TEAMVALLEYTRADING
ESTATE,
GATESHEAD-ON-TYNE.

[Received, November 5th, 1942.j